Everpure accelerates AI workloads with Data Stream and unveils data-primacy architectural vision

Big-data storage company Everpure Inc., formerly known as Pure Storage, is rethinking enterprise data architectures to facilitate better access and scalability for artificial intelligence workloads.

At its annual customer conference, Pure Accelerate, the company today announced the immediate availability of Everpure Data Stream, which helps to transform raw, unstructured data so enterprises can feed it into AI models more easily.

Alongside Data Stream, the company unveiled Everpure Data Intelligence to simplify data discovery and classification, plus significant performance upgrades to its Enterprise Data Cloud platform. The announcements signal a fundamental shift in strategy for Everpure, which is urging customers to abandon their old application-centric data architectures in favor of a new “data primacy” model.

Everpure Chairman and Chief Executive Charles Giancarlo introduced the new offerings in a blog post, where he explained that the biggest hurdle for enterprise AI is no longer the models, but the complexity of the data pipelines needed to feed them with information. He cites a recent survey by International Data Corp., which found that 94% of information technology leaders consider data quality to be the ultimate factor when it comes to getting value from AI.

With Everpure Data Stream, the company is tackling the problem of data access with a new engine that automates raw data preparation in order to reduce ingestion times from months to a matter of minutes. At the same time, the new platform will enforce access controls at the data stream level to ensure proprietary data remains secure.

The company explained that Data Stream builds on Nvidia Corp.’s AI Data Platform reference design, replacing traditional and heavily manual data ingestion processes with a graphics processing unit-accelerated pipeline that extends all the way to the inference stage. Giancarlo said this architecture allows both compute and storage to scale independently of one another, ensuring that GPU clusters will never be starved of data, solving a key problem around compute efficiency. It’s enabled by Everpure’s FlashBlade storage systems, including FlashBlade//S and FlashBlade/EXA, which Giancarlo said can push through data at an extremely rapid throughput of 800 gigabytes per node, eliminating GPU idling once and for all.

The move is aligned with a broader storage industry shift, with all of the major players looking to focus more on the data and less about where the bits live, Steve McDowell of NAND Research told SiliconANGLE. He explained that NetApp Inc.’s Intelligent Data Infrastructure and Dell’s AI Data Platform are earlier examples of this trend, which is all about ensuring AI gets enough data to keep up its relentless pace of work. “Pure put a big stake in the ground with its rebrand earlier this year,” McDowell said. “Today’s announcements are the first real products aligned with its new mission.

Everpure Chief Technology Officer Robert Lee said future enterprise data architectures will require a unified platform that enables businesses to begin with smaller projects before scaling up to handle exabytes of data. “Everpure solves this challenge by delivering a trusted, secure and high-performance data pipeline that accelerates time-to-results for an enterprise’s data,” he said.

The data-primacy architecture

Pipeline performance is only one part of the equation when it comes to streamlining AI data access, which is why Everpure also debuted a new technology called Everpure Data Intelligence. It’s based on the capabilities of a company called 1touch.io Inc., which was acquired just over three months ago when Everpure announced its rebrand.

1touch.io sold a platform called Kontxtual, which is used to scan enterprise’s on-premises data infrastructures and cloud environments to create a comprehensive inventory of their data assets, along with descriptions of what each one is.

The Everpure Data Intelligence layer is meant to address the semantic fragmentation of enterprise data by discovering, classifying and contextualizing both structured and unstructured information at its source. It works by mapping dependencies into a universal semantic knowledge graph, which autonomous AI agents can then access via the Model Context Protocol to obtain real-time context about whatever it is they’re working on. According to Giancarlo, this has the effect of minimizing AI tokens costs while reducing compliance risks, because it also applies automated governance to protect sensitive assets such as personally identifiable information.

McDowell believes that Everpure Data Intelligence is going to be the most impactful of today’s announcements. “It instantly gives the company data awareness capabilities, along with the automated data ingest provided by Data Stream, that equal or exceed the capabilities of its rivals,” he explained. “1touch.io was a strong acquisition by Everpure and it’s yielding immediate returns. Data Intelligence puts the company on a solid foundation for sovereign AI data needs, while also adding the contextual detail that’s so important for agentic AI workflows.”

Everpure announced a series of updates to the Enterprise Data Cloud platform too. For instance, it’s getting a boost with Evergreen//One Overdrive, which will launch in the third quarter and provide a 25% performance boost for on-premises storage during traffic spikes. There’s also a new Intelligent Control Plane that enables natural language copilot workflow execution, better cyber anomaly detection and automatic workload rebalancing. When this launches later in the year, it will enable companies to replace reactive storage management to a self-optimized system, the company said. Last but not least, it introduced support for native virtual machines on Microsoft Azure.

The latter is an especially interesting move by Everpure, because it appears to be taking a much more measured approach than one of its chief competitors, NetApp, which has built up a strong presence in cloud storage due to the availability of its ONTAP storage service on public cloud platforms, McDowell said. “The Azure-native VMs allow Everpure’s hybrid cloud customers to move data-heavy workloads to Azure without refactoring the underlying storage, eliminating the pain of migrating data,” the analyst explained. “It’s a powerful capability for those enterprises that need it.”

Giancarlo said the emergence of AI underscores his belief that software applications should not be the center of gravity of IT architectures. Most companies have embraced an application-centric design that has resulted in massive app sprawl with siloed definitions of data and slow and costly extraction process that lead to inferior AI performance.

“AI completely upends the traditional IT hierarchy; enterprises that do not shift from app-centricity to data primacy will fall behind,” he argued. “Because data is a company’s primary asset, embedding context, semantics and governance directly at the data layer is the right way to reduce data fragmentation.”

Giancarlo believes that the transition to a data-primacy framework should become a major priority for enterprises. “A company’s data is now a growing corpus of intelligence whose strategic value far outweighs the applications that created it,” he added. “Everyone talks about the importance of data. Now it’s time to put data first, and everything else will follow.”

Image: Pure Storage